lib/json.c

   1 /*
   2  * Copyright (c) 2009, 2010 Nicira Networks.
   3  *
   4  * Licensed under the Apache License, Version 2.0 (the "License");
   5  * you may not use this file except in compliance with the License.
   6  * You may obtain a copy of the License at:
   7  *
   8  *     http://www.apache.org/licenses/LICENSE-2.0
   9  *
  10  * Unless required by applicable law or agreed to in writing, software
  11  * distributed under the License is distributed on an "AS IS" BASIS,
  12  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  13  * See the License for the specific language governing permissions and
  14  * limitations under the License.
  15  */
  16
  17 #include <config.h>
  18
  19 #include "json.h"
  20
  21 #include <assert.h>
  22 #include <ctype.h>
  23 #include <errno.h>
  24 #include <float.h>
  25 #include <limits.h>
  26 #include <math.h>
  27 #include <string.h>
  28
  29 #include "dynamic-string.h"
  30 #include "hash.h"
  31 #include "shash.h"
  32 #include "unicode.h"
  33 #include "util.h"
  34
  35 /* The type of a JSON token. */
  36 enum json_token_type {
  37     T_EOF = 0,
  38     T_BEGIN_ARRAY = '[',
  39     T_END_ARRAY = ']',
  40     T_BEGIN_OBJECT = '{',
  41     T_END_OBJECT = '}',
  42     T_NAME_SEPARATOR = ':',
  43     T_VALUE_SEPARATOR = ',',
  44     T_FALSE = UCHAR_MAX + 1,
  45     T_NULL,
  46     T_TRUE,
  47     T_INTEGER,
  48     T_REAL,
  49     T_STRING
  50 };
  51
  52 /* A JSON token.
  53  *
  54  * RFC 4627 doesn't define a lexical structure for JSON but I believe this to
  55  * be compliant with the standard.
  56  */
  57 struct json_token {
  58     enum json_token_type type;
  59     union {
  60         double real;
  61         long long int integer;
  62         const char *string;
  63     } u;
  64 };
  65
  66 enum json_lex_state {
  67     JSON_LEX_START,             /* Not inside a token. */
  68     JSON_LEX_NUMBER,            /* Reading a number. */
  69     JSON_LEX_KEYWORD,           /* Reading a keyword. */
  70     JSON_LEX_STRING,            /* Reading a quoted string. */
  71     JSON_LEX_ESCAPE             /* In a quoted string just after a "\". */
  72 };
  73
  74 enum json_parse_state {
  75     JSON_PARSE_START,           /* Beginning of input. */
  76     JSON_PARSE_END,             /* End of input. */
  77
  78     /* Objects. */
  79     JSON_PARSE_OBJECT_INIT,     /* Expecting '}' or an object name. */
  80     JSON_PARSE_OBJECT_NAME,     /* Expecting an object name. */
  81     JSON_PARSE_OBJECT_COLON,    /* Expecting ':'. */
  82     JSON_PARSE_OBJECT_VALUE,    /* Expecting an object value. */
  83     JSON_PARSE_OBJECT_NEXT,     /* Expecting ',' or '}'. */
  84
  85     /* Arrays. */
  86     JSON_PARSE_ARRAY_INIT,      /* Expecting ']' or a value. */
  87     JSON_PARSE_ARRAY_VALUE,     /* Expecting a value. */
  88     JSON_PARSE_ARRAY_NEXT       /* Expecting ',' or ']'. */
  89 };
  90
  91 struct json_parser_node {
  92     struct json *json;
  93 };
  94
  95 /* A JSON parser. */
  96 struct json_parser {
  97     int flags;
  98
  99     /* Lexical analysis. */
 100     enum json_lex_state lex_state;
 101     struct ds buffer;           /* Buffer for accumulating token text. */
 102     int line_number;
 103     int column_number;
 104     int byte_number;
 105
 106     /* Parsing. */
 107     enum json_parse_state parse_state;
 108 #define JSON_MAX_HEIGHT 1000
 109     struct json_parser_node *stack;
 110     size_t height, allocated_height;
 111     char *member_name;
 112
 113     /* Parse status. */
 114     bool done;
 115     char *error;                /* Error message, if any, null if none yet. */
 116 };
 117
 118 static struct json *json_create(enum json_type type);
 119 static void json_parser_input(struct json_parser *, struct json_token *);
 120
 121 static void json_error(struct json_parser *p, const char *format, ...)
 122     PRINTF_FORMAT(2, 3);
 123 \f
 124 const char *
 125 json_type_to_string(enum json_type type)
 126 {
 127     switch (type) {
 128     case JSON_NULL:
 129         return "null";
 130
 131     case JSON_FALSE:
 132         return "false";
 133
 134     case JSON_TRUE:
 135         return "true";
 136
 137     case JSON_OBJECT:
 138         return "object";
 139
 140     case JSON_ARRAY:
 141         return "array";
 142
 143     case JSON_INTEGER:
 144     case JSON_REAL:
 145         return "number";
 146
 147     case JSON_STRING:
 148         return "string";
 149
 150     case JSON_N_TYPES:
 151     default:
 152         return "<invalid>";
 153     }
 154 }
 155 \f
 156 /* Functions for manipulating struct json. */
 157
 158 struct json *
 159 json_null_create(void)
 160 {
 161     return json_create(JSON_NULL);
 162 }
 163
 164 struct json *
 165 json_boolean_create(bool b)
 166 {
 167     return json_create(b ? JSON_TRUE : JSON_FALSE);
 168 }
 169
 170 struct json *
 171 json_string_create_nocopy(char *s)
 172 {
 173     struct json *json = json_create(JSON_STRING);
 174     json->u.string = s;
 175     return json;
 176 }
 177
 178 struct json *
 179 json_string_create(const char *s)
 180 {
 181     return json_string_create_nocopy(xstrdup(s));
 182 }
 183
 184 struct json *
 185 json_array_create_empty(void)
 186 {
 187     struct json *json = json_create(JSON_ARRAY);
 188     json->u.array.elems = NULL;
 189     json->u.array.n = 0;
 190     json->u.array.n_allocated = 0;
 191     return json;
 192 }
 193
 194 void
 195 json_array_add(struct json *array_, struct json *element)
 196 {
 197     struct json_array *array = json_array(array_);
 198     if (array->n >= array->n_allocated) {
 199         array->elems = x2nrealloc(array->elems, &array->n_allocated,
 200                                   sizeof *array->elems);
 201     }
 202     array->elems[array->n++] = element;
 203 }
 204
 205 void
 206 json_array_trim(struct json *array_)
 207 {
 208     struct json_array *array = json_array(array_);
 209     if (array->n < array->n_allocated){
 210         array->n_allocated = array->n;
 211         array->elems = xrealloc(array->elems, array->n * sizeof *array->elems);
 212     }
 213 }
 214
 215 struct json *
 216 json_array_create(struct json **elements, size_t n)
 217 {
 218     struct json *json = json_create(JSON_ARRAY);
 219     json->u.array.elems = elements;
 220     json->u.array.n = n;
 221     json->u.array.n_allocated = n;
 222     return json;
 223 }
 224
 225 struct json *
 226 json_array_create_1(struct json *elem0)
 227 {
 228     struct json **elems = xmalloc(sizeof *elems);
 229     elems[0] = elem0;
 230     return json_array_create(elems, 1);
 231 }
 232
 233 struct json *
 234 json_array_create_2(struct json *elem0, struct json *elem1)
 235 {
 236     struct json **elems = xmalloc(2 * sizeof *elems);
 237     elems[0] = elem0;
 238     elems[1] = elem1;
 239     return json_array_create(elems, 2);
 240 }
 241
 242 struct json *
 243 json_array_create_3(struct json *elem0, struct json *elem1, struct json *elem2)
 244 {
 245     struct json **elems = xmalloc(3 * sizeof *elems);
 246     elems[0] = elem0;
 247     elems[1] = elem1;
 248     elems[2] = elem2;
 249     return json_array_create(elems, 3);
 250 }
 251
 252 struct json *
 253 json_object_create(void)
 254 {
 255     struct json *json = json_create(JSON_OBJECT);
 256     json->u.object = xmalloc(sizeof *json->u.object);
 257     shash_init(json->u.object);
 258     return json;
 259 }
 260
 261 struct json *
 262 json_integer_create(long long int integer)
 263 {
 264     struct json *json = json_create(JSON_INTEGER);
 265     json->u.integer = integer;
 266     return json;
 267 }
 268
 269 struct json *
 270 json_real_create(double real)
 271 {
 272     struct json *json = json_create(JSON_REAL);
 273     json->u.real = real;
 274     return json;
 275 }
 276
 277 void
 278 json_object_put(struct json *json, const char *name, struct json *value)
 279 {
 280     json_destroy(shash_replace(json->u.object, name, value));
 281 }
 282
 283 void
 284 json_object_put_string(struct json *json, const char *name, const char *value)
 285 {
 286     json_object_put(json, name, json_string_create(value));
 287 }
 288
 289 const char *
 290 json_string(const struct json *json)
 291 {
 292     assert(json->type == JSON_STRING);
 293     return json->u.string;
 294 }
 295
 296 struct json_array *
 297 json_array(const struct json *json)
 298 {
 299     assert(json->type == JSON_ARRAY);
 300     return (struct json_array *) &json->u.array;
 301 }
 302
 303 struct shash *
 304 json_object(const struct json *json)
 305 {
 306     assert(json->type == JSON_OBJECT);
 307     return (struct shash *) json->u.object;
 308 }
 309
 310 bool
 311 json_boolean(const struct json *json)
 312 {
 313     assert(json->type == JSON_TRUE || json->type == JSON_FALSE);
 314     return json->type == JSON_TRUE;
 315 }
 316
 317 double
 318 json_real(const struct json *json)
 319 {
 320     assert(json->type == JSON_REAL || json->type == JSON_INTEGER);
 321     return json->type == JSON_REAL ? json->u.real : json->u.integer;
 322 }
 323
 324 int64_t
 325 json_integer(const struct json *json)
 326 {
 327     assert(json->type == JSON_INTEGER);
 328     return json->u.integer;
 329 }
 330 \f
 331 static void json_destroy_object(struct shash *object);
 332 static void json_destroy_array(struct json_array *array);
 333
 334 /* Frees 'json' and everything it points to, recursively. */
 335 void
 336 json_destroy(struct json *json)
 337 {
 338     if (json) {
 339         switch (json->type) {
 340         case JSON_OBJECT:
 341             json_destroy_object(json->u.object);
 342             break;
 343
 344         case JSON_ARRAY:
 345             json_destroy_array(&json->u.array);
 346             break;
 347
 348         case JSON_STRING:
 349             free(json->u.string);
 350             break;
 351
 352         case JSON_NULL:
 353         case JSON_FALSE:
 354         case JSON_TRUE:
 355         case JSON_INTEGER:
 356         case JSON_REAL:
 357             break;
 358
 359         case JSON_N_TYPES:
 360             NOT_REACHED();
 361         }
 362         free(json);
 363     }
 364 }
 365
 366 static void
 367 json_destroy_object(struct shash *object)
 368 {
 369     struct shash_node *node, *next;
 370
 371     SHASH_FOR_EACH_SAFE (node, next, object) {
 372         struct json *value = node->data;
 373
 374         json_destroy(value);
 375         shash_delete(object, node);
 376     }
 377     shash_destroy(object);
 378     free(object);
 379 }
 380
 381 static void
 382 json_destroy_array(struct json_array *array)
 383 {
 384     size_t i;
 385
 386     for (i = 0; i < array->n; i++) {
 387         json_destroy(array->elems[i]);
 388     }
 389     free(array->elems);
 390 }
 391 \f
 392 static struct json *json_clone_object(const struct shash *object);
 393 static struct json *json_clone_array(const struct json_array *array);
 394
 395 /* Returns a deep copy of 'json'. */
 396 struct json *
 397 json_clone(const struct json *json)
 398 {
 399     switch (json->type) {
 400     case JSON_OBJECT:
 401         return json_clone_object(json->u.object);
 402
 403     case JSON_ARRAY:
 404         return json_clone_array(&json->u.array);
 405
 406     case JSON_STRING:
 407         return json_string_create(json->u.string);
 408
 409     case JSON_NULL:
 410     case JSON_FALSE:
 411     case JSON_TRUE:
 412         return json_create(json->type);
 413
 414     case JSON_INTEGER:
 415         return json_integer_create(json->u.integer);
 416
 417     case JSON_REAL:
 418         return json_real_create(json->u.real);
 419
 420     case JSON_N_TYPES:
 421     default:
 422         NOT_REACHED();
 423     }
 424 }
 425
 426 static struct json *
 427 json_clone_object(const struct shash *object)
 428 {
 429     struct shash_node *node;
 430     struct json *json;
 431
 432     json = json_object_create();
 433     SHASH_FOR_EACH (node, object) {
 434         struct json *value = node->data;
 435         json_object_put(json, node->name, json_clone(value));
 436     }
 437     return json;
 438 }
 439
 440 static struct json *
 441 json_clone_array(const struct json_array *array)
 442 {
 443     struct json **elems;
 444     size_t i;
 445
 446     elems = xmalloc(array->n * sizeof *elems);
 447     for (i = 0; i < array->n; i++) {
 448         elems[i] = json_clone(array->elems[i]);
 449     }
 450     return json_array_create(elems, array->n);
 451 }
 452 \f
 453 static size_t
 454 json_hash_object(const struct shash *object, size_t basis)
 455 {
 456     const struct shash_node **nodes;
 457     size_t n, i;
 458
 459     nodes = shash_sort(object);
 460     n = shash_count(object);
 461     for (i = 0; i < n; i++) {
 462         const struct shash_node *node = nodes[i];
 463         basis = hash_string(node->name, basis);
 464         basis = json_hash(node->data, basis);
 465     }
 466     return basis;
 467 }
 468
 469 static size_t
 470 json_hash_array(const struct json_array *array, size_t basis)
 471 {
 472     size_t i;
 473
 474     basis = hash_int(array->n, basis);
 475     for (i = 0; i < array->n; i++) {
 476         basis = json_hash(array->elems[i], basis);
 477     }
 478     return basis;
 479 }
 480
 481 size_t
 482 json_hash(const struct json *json, size_t basis)
 483 {
 484     switch (json->type) {
 485     case JSON_OBJECT:
 486         return json_hash_object(json->u.object, basis);
 487
 488     case JSON_ARRAY:
 489         return json_hash_array(&json->u.array, basis);
 490
 491     case JSON_STRING:
 492         return hash_string(json->u.string, basis);
 493
 494     case JSON_NULL:
 495     case JSON_FALSE:
 496     case JSON_TRUE:
 497         return hash_int(json->type << 8, basis);
 498
 499     case JSON_INTEGER:
 500         return hash_int(json->u.integer, basis);
 501
 502     case JSON_REAL:
 503         return hash_double(json->u.real, basis);
 504
 505     case JSON_N_TYPES:
 506     default:
 507         NOT_REACHED();
 508     }
 509 }
 510
 511 static bool
 512 json_equal_object(const struct shash *a, const struct shash *b)
 513 {
 514     struct shash_node *a_node;
 515
 516     if (shash_count(a) != shash_count(b)) {
 517         return false;
 518     }
 519
 520     SHASH_FOR_EACH (a_node, a) {
 521         struct shash_node *b_node = shash_find(b, a_node->name);
 522         if (!b_node || !json_equal(a_node->data, b_node->data)) {
 523             return false;
 524         }
 525     }
 526
 527     return true;
 528 }
 529
 530 static bool
 531 json_equal_array(const struct json_array *a, const struct json_array *b)
 532 {
 533     size_t i;
 534
 535     if (a->n != b->n) {
 536         return false;
 537     }
 538
 539     for (i = 0; i < a->n; i++) {
 540         if (!json_equal(a->elems[i], b->elems[i])) {
 541             return false;
 542         }
 543     }
 544
 545     return true;
 546 }
 547
 548 bool
 549 json_equal(const struct json *a, const struct json *b)
 550 {
 551     if (a->type != b->type) {
 552         return false;
 553     }
 554
 555     switch (a->type) {
 556     case JSON_OBJECT:
 557         return json_equal_object(a->u.object, b->u.object);
 558
 559     case JSON_ARRAY:
 560         return json_equal_array(&a->u.array, &b->u.array);
 561
 562     case JSON_STRING:
 563         return !strcmp(a->u.string, b->u.string);
 564
 565     case JSON_NULL:
 566     case JSON_FALSE:
 567     case JSON_TRUE:
 568         return true;
 569
 570     case JSON_INTEGER:
 571         return a->u.integer == b->u.integer;
 572
 573     case JSON_REAL:
 574         return a->u.real == b->u.real;
 575
 576     case JSON_N_TYPES:
 577     default:
 578         NOT_REACHED();
 579     }
 580 }
 581 \f
 582 /* Lexical analysis. */
 583
 584 static void
 585 json_lex_keyword(struct json_parser *p)
 586 {
 587     struct json_token token;
 588     const char *s;
 589
 590     s = ds_cstr(&p->buffer);
 591     if (!strcmp(s, "false")) {
 592         token.type = T_FALSE;
 593     } else if (!strcmp(s, "true")) {
 594         token.type = T_TRUE;
 595     } else if (!strcmp(s, "null")) {
 596         token.type = T_NULL;
 597     } else {
 598         json_error(p, "invalid keyword '%s'", s);
 599         return;
 600     }
 601     json_parser_input(p, &token);
 602 }
 603
 604 static void
 605 json_lex_number(struct json_parser *p)
 606 {
 607     const char *cp = ds_cstr(&p->buffer);
 608     unsigned long long int significand = 0;
 609     struct json_token token;
 610     bool imprecise = false;
 611     bool negative = false;
 612     int pow10 = 0;
 613
 614     /* Leading minus sign. */
 615     if (*cp == '-') {
 616         negative = true;
 617         cp++;
 618     }
 619
 620     /* At least one integer digit, but 0 may not be used as a leading digit for
 621      * a longer number. */
 622     significand = 0;
 623     if (*cp == '0') {
 624         cp++;
 625         if (isdigit(*cp)) {
 626             json_error(p, "leading zeros not allowed");
 627             return;
 628         }
 629     } else if (isdigit(*cp)) {
 630         do {
 631             if (significand <= ULLONG_MAX / 10) {
 632                 significand = significand * 10 + (*cp - '0');
 633             } else {
 634                 pow10++;
 635                 if (*cp != '0') {
 636                     imprecise = true;
 637                 }
 638             }
 639             cp++;
 640         } while (isdigit(*cp));
 641     } else {
 642         json_error(p, "'-' must be followed by digit");
 643         return;
 644     }
 645
 646     /* Optional fraction. */
 647     if (*cp == '.') {
 648         cp++;
 649         if (!isdigit(*cp)) {
 650             json_error(p, "decimal point must be followed by digit");
 651             return;
 652         }
 653         do {
 654             if (significand <= ULLONG_MAX / 10) {
 655                 significand = significand * 10 + (*cp - '0');
 656                 pow10--;
 657             } else if (*cp != '0') {
 658                 imprecise = true;
 659             }
 660             cp++;
 661         } while (isdigit(*cp));
 662     }
 663
 664     /* Optional exponent. */
 665     if (*cp == 'e' || *cp == 'E') {
 666         bool negative_exponent = false;
 667         int exponent;
 668
 669         cp++;
 670         if (*cp == '+') {
 671             cp++;
 672         } else if (*cp == '-') {
 673             negative_exponent = true;
 674             cp++;
 675         }
 676
 677         if (!isdigit(*cp)) {
 678             json_error(p, "exponent must contain at least one digit");
 679             return;
 680         }
 681
 682         exponent = 0;
 683         do {
 684             if (exponent >= INT_MAX / 10) {
 685                 json_error(p, "exponent outside valid range");
 686                 return;
 687             }
 688             exponent = exponent * 10 + (*cp - '0');
 689             cp++;
 690         } while (isdigit(*cp));
 691
 692         if (negative_exponent) {
 693             pow10 -= exponent;
 694         } else {
 695             pow10 += exponent;
 696         }
 697     }
 698
 699     if (*cp != '\0') {
 700         json_error(p, "syntax error in number");
 701         return;
 702     }
 703
 704     /* Figure out number.
 705      *
 706      * We suppress negative zeros as a matter of policy. */
 707     if (!significand) {
 708         struct json_token token;
 709         token.type = T_INTEGER;
 710         token.u.integer = 0;
 711         json_parser_input(p, &token);
 712         return;
 713     }
 714
 715     if (!imprecise) {
 716         while (pow10 > 0 && significand < ULLONG_MAX / 10) {
 717             significand *= 10;
 718             pow10--;
 719         }
 720         while (pow10 < 0 && significand % 10 == 0) {
 721             significand /= 10;
 722             pow10++;
 723         }
 724         if (pow10 == 0
 725             && significand <= (negative
 726                                ? (unsigned long long int) LLONG_MAX + 1
 727                                : LLONG_MAX)) {
 728             token.type = T_INTEGER;
 729             token.u.integer = negative ? -significand : significand;
 730             json_parser_input(p, &token);
 731             return;
 732         }
 733     }
 734
 735     token.type = T_REAL;
 736     if (!str_to_double(ds_cstr(&p->buffer), &token.u.real)) {
 737         json_error(p, "number outside valid range");
 738         return;
 739     }
 740     /* Suppress negative zero. */
 741     if (token.u.real == 0) {
 742         token.u.real = 0;
 743     }
 744     json_parser_input(p, &token);
 745 }
 746
 747 static const char *
 748 json_lex_4hex(const char *cp, const char *end, int *valuep)
 749 {
 750     int value, i;
 751
 752     if (cp + 4 > end) {
 753         return "quoted string ends within \\u escape";
 754     }
 755
 756     value = 0;
 757     for (i = 0; i < 4; i++) {
 758         unsigned char c = *cp++;
 759         if (!isxdigit(c)) {
 760             return "malformed \\u escape";
 761         }
 762         value = (value << 4) | hexit_value(c);
 763     }
 764     if (!value) {
 765         return "null bytes not supported in quoted strings";
 766     }
 767     *valuep = value;
 768     return NULL;
 769 }
 770
 771 static const char *
 772 json_lex_unicode(const char *cp, const char *end, struct ds *out)
 773 {
 774     const char *error;
 775     int c0, c1;
 776
 777     error = json_lex_4hex(cp, end, &c0);
 778     if (error) {
 779         ds_clear(out);
 780         ds_put_cstr(out, error);
 781         return NULL;
 782     }
 783     cp += 4;
 784     if (!uc_is_leading_surrogate(c0)) {
 785         ds_put_utf8(out, c0);
 786         return cp;
 787     }
 788
 789     if (cp + 2 > end || *cp++ != '\\' || *cp++ != 'u') {
 790         ds_clear(out);
 791         ds_put_cstr(out, "malformed escaped surrogate pair");
 792         return NULL;
 793     }
 794
 795     error = json_lex_4hex(cp, end, &c1);
 796     if (error) {
 797         ds_clear(out);
 798         ds_put_cstr(out, error);
 799         return NULL;
 800     }
 801     cp += 4;
 802     if (!uc_is_trailing_surrogate(c1)) {
 803         ds_clear(out);
 804         ds_put_cstr(out, "second half of escaped surrogate pair is not "
 805                     "trailing surrogate");
 806         return NULL;
 807     }
 808
 809     ds_put_utf8(out, utf16_decode_surrogate_pair(c0, c1));
 810     return cp;
 811 }
 812
 813 bool
 814 json_string_unescape(const char *in, size_t in_len, char **outp)
 815 {
 816     const char *end = in + in_len;
 817     bool ok = false;
 818     struct ds out;
 819
 820     ds_init(&out);
 821     ds_reserve(&out, in_len);
 822     if (in_len > 0 && in[in_len - 1] == '\\') {
 823         ds_put_cstr(&out, "quoted string may not end with backslash");
 824         goto exit;
 825     }
 826     while (in < end) {
 827         if (*in == '"') {
 828             ds_clear(&out);
 829             ds_put_cstr(&out, "quoted string may not include unescaped \"");
 830             goto exit;
 831         }
 832         if (*in != '\\') {
 833             ds_put_char(&out, *in++);
 834             continue;
 835         }
 836
 837         in++;
 838         switch (*in++) {
 839         case '"': case '\\': case '/':
 840             ds_put_char(&out, in[-1]);
 841             break;
 842
 843         case 'b':
 844             ds_put_char(&out, '\b');
 845             break;
 846
 847         case 'f':
 848             ds_put_char(&out, '\f');
 849             break;
 850
 851         case 'n':
 852             ds_put_char(&out, '\n');
 853             break;
 854
 855         case 'r':
 856             ds_put_char(&out, '\r');
 857             break;
 858
 859         case 't':
 860             ds_put_char(&out, '\t');
 861             break;
 862
 863         case 'u':
 864             in = json_lex_unicode(in, end, &out);
 865             if (!in) {
 866                 goto exit;
 867             }
 868             break;
 869
 870         default:
 871             ds_clear(&out);
 872             ds_put_format(&out, "bad escape \\%c", in[-1]);
 873             goto exit;
 874         }
 875     }
 876     ok = true;
 877
 878 exit:
 879     *outp = ds_cstr(&out);
 880     return ok;
 881 }
 882
 883 static void
 884 json_parser_input_string(struct json_parser *p, const char *s)
 885 {
 886     struct json_token token;
 887
 888     token.type = T_STRING;
 889     token.u.string = s;
 890     json_parser_input(p, &token);
 891 }
 892
 893 static void
 894 json_lex_string(struct json_parser *p)
 895 {
 896     const char *raw = ds_cstr(&p->buffer);
 897     if (!strchr(raw, '\\')) {
 898         json_parser_input_string(p, raw);
 899     } else {
 900         char *cooked;
 901
 902         if (json_string_unescape(raw, strlen(raw), &cooked)) {
 903             json_parser_input_string(p, cooked);
 904         } else {
 905             json_error(p, "%s", cooked);
 906         }
 907
 908         free(cooked);
 909     }
 910 }
 911
 912 static bool
 913 json_lex_input(struct json_parser *p, unsigned char c)
 914 {
 915     struct json_token token;
 916
 917     p->byte_number++;
 918     if (c == '\n') {
 919         p->column_number = 0;
 920         p->line_number++;
 921     } else {
 922         p->column_number++;
 923     }
 924
 925     switch (p->lex_state) {
 926     case JSON_LEX_START:
 927         switch (c) {
 928         case ' ': case '\t': case '\n': case '\r':
 929             /* Nothing to do. */
 930             return true;
 931
 932         case 'a': case 'b': case 'c': case 'd': case 'e':
 933         case 'f': case 'g': case 'h': case 'i': case 'j':
 934         case 'k': case 'l': case 'm': case 'n': case 'o':
 935         case 'p': case 'q': case 'r': case 's': case 't':
 936         case 'u': case 'v': case 'w': case 'x': case 'y':
 937         case 'z':
 938             p->lex_state = JSON_LEX_KEYWORD;
 939             break;
 940
 941         case '[': case '{': case ']': case '}': case ':': case ',':
 942             token.type = c;
 943             json_parser_input(p, &token);
 944             return true;
 945
 946         case '-':
 947         case '0': case '1': case '2': case '3': case '4':
 948         case '5': case '6': case '7': case '8': case '9':
 949             p->lex_state = JSON_LEX_NUMBER;
 950             break;
 951
 952         case '"':
 953             p->lex_state = JSON_LEX_STRING;
 954             return true;
 955
 956         default:
 957             if (isprint(c)) {
 958                 json_error(p, "invalid character '%c'", c);
 959             } else {
 960                 json_error(p, "invalid character U+%04x", c);
 961             }
 962             return true;
 963         }
 964         break;
 965
 966     case JSON_LEX_KEYWORD:
 967         if (!isalpha((unsigned char) c)) {
 968             json_lex_keyword(p);
 969             return false;
 970         }
 971         break;
 972
 973     case JSON_LEX_NUMBER:
 974         if (!strchr(".0123456789eE-+", c)) {
 975             json_lex_number(p);
 976             return false;
 977         }
 978         break;
 979
 980     case JSON_LEX_STRING:
 981         if (c == '\\') {
 982             p->lex_state = JSON_LEX_ESCAPE;
 983         } else if (c == '"') {
 984             json_lex_string(p);
 985             return true;
 986         } else if (c < 0x20) {
 987             json_error(p, "U+%04X must be escaped in quoted string", c);
 988             return true;
 989         }
 990         break;
 991
 992     case JSON_LEX_ESCAPE:
 993         p->lex_state = JSON_LEX_STRING;
 994         break;
 995
 996     default:
 997         abort();
 998     }
 999     ds_put_char(&p->buffer, c);
1000     return true;
1001 }
1002 \f
1003 /* Parsing. */
1004
1005 /* Parses 'string' as a JSON object or array and returns a newly allocated
1006  * 'struct json'.  The caller must free the returned structure with
1007  * json_destroy() when it is no longer needed.
1008  *
1009  * 'string' must be encoded in UTF-8.
1010  *
1011  * If 'string' is valid JSON, then the returned 'struct json' will be either an
1012  * object (JSON_OBJECT) or an array (JSON_ARRAY).
1013  *
1014  * If 'string' is not valid JSON, then the returned 'struct json' will be a
1015  * string (JSON_STRING) that describes the particular error encountered during
1016  * parsing.  (This is an acceptable means of error reporting because at its top
1017  * level JSON must be either an object or an array; a bare string is not
1018  * valid.) */
1019 struct json *
1020 json_from_string(const char *string)
1021 {
1022     struct json_parser *p = json_parser_create(JSPF_TRAILER);
1023     json_parser_feed(p, string, strlen(string));
1024     return json_parser_finish(p);
1025 }
1026
1027 /* Reads the file named 'file_name', parses its contents as a JSON object or
1028  * array, and returns a newly allocated 'struct json'.  The caller must free
1029  * the returned structure with json_destroy() when it is no longer needed.
1030  *
1031  * The file must be encoded in UTF-8.
1032  *
1033  * See json_from_string() for return value semantics.
1034  */
1035 struct json *
1036 json_from_file(const char *file_name)
1037 {
1038     struct json *json;
1039     FILE *stream;
1040
1041     stream = fopen(file_name, "r");
1042     if (!stream) {
1043         return json_string_create_nocopy(
1044             xasprintf("error opening \"%s\": %s", file_name, strerror(errno)));
1045     }
1046     json = json_from_stream(stream);
1047     fclose(stream);
1048
1049     return json;
1050 }
1051
1052 /* Parses the contents of 'stream' as a JSON object or array, and returns a
1053  * newly allocated 'struct json'.  The caller must free the returned structure
1054  * with json_destroy() when it is no longer needed.
1055  *
1056  * The file must be encoded in UTF-8.
1057  *
1058  * See json_from_string() for return value semantics.
1059  */
1060 struct json *
1061 json_from_stream(FILE *stream)
1062 {
1063     struct json_parser *p;
1064     struct json *json;
1065
1066     p = json_parser_create(JSPF_TRAILER);
1067     for (;;) {
1068         char buffer[BUFSIZ];
1069         size_t n;
1070
1071         n = fread(buffer, 1, sizeof buffer, stream);
1072         if (!n || json_parser_feed(p, buffer, n) != n) {
1073             break;
1074         }
1075     }
1076     json = json_parser_finish(p);
1077
1078     if (ferror(stream)) {
1079         json_destroy(json);
1080         json = json_string_create_nocopy(
1081             xasprintf("error reading JSON stream: %s", strerror(errno)));
1082     }
1083
1084     return json;
1085 }
1086
1087 struct json_parser *
1088 json_parser_create(int flags)
1089 {
1090     struct json_parser *p = xzalloc(sizeof *p);
1091     p->flags = flags;
1092     return p;
1093 }
1094
1095 size_t
1096 json_parser_feed(struct json_parser *p, const char *input, size_t n)
1097 {
1098     size_t i;
1099     for (i = 0; !p->done && i < n; ) {
1100         if (json_lex_input(p, input[i])) {
1101             i++;
1102         }
1103     }
1104     return i;
1105 }
1106
1107 bool
1108 json_parser_is_done(const struct json_parser *p)
1109 {
1110     return p->done;
1111 }
1112
1113 struct json *
1114 json_parser_finish(struct json_parser *p)
1115 {
1116     struct json *json;
1117
1118     switch (p->lex_state) {
1119     case JSON_LEX_START:
1120         break;
1121
1122     case JSON_LEX_STRING:
1123     case JSON_LEX_ESCAPE:
1124         json_error(p, "unexpected end of input in quoted string");
1125         break;
1126
1127     case JSON_LEX_NUMBER:
1128     case JSON_LEX_KEYWORD:
1129         json_lex_input(p, ' ');
1130         break;
1131     }
1132
1133     if (p->parse_state == JSON_PARSE_START) {
1134         json_error(p, "empty input stream");
1135     } else if (p->parse_state != JSON_PARSE_END) {
1136         json_error(p, "unexpected end of input");
1137     }
1138
1139     if (!p->error) {
1140         assert(p->height == 1);
1141         assert(p->stack[0].json != NULL);
1142         json = p->stack[--p->height].json;
1143     } else {
1144         json = json_string_create_nocopy(p->error);
1145         p->error = NULL;
1146     }
1147
1148     json_parser_abort(p);
1149
1150     return json;
1151 }
1152
1153 void
1154 json_parser_abort(struct json_parser *p)
1155 {
1156     if (p) {
1157         ds_destroy(&p->buffer);
1158         if (p->height) {
1159             json_destroy(p->stack[0].json);
1160         }
1161         free(p->stack);
1162         free(p->member_name);
1163         free(p->error);
1164         free(p);
1165     }
1166 }
1167
1168 static struct json_parser_node *
1169 json_parser_top(struct json_parser *p)
1170 {
1171     return &p->stack[p->height - 1];
1172 }
1173
1174 static void
1175 json_parser_put_value(struct json_parser *p, struct json *value)
1176 {
1177     struct json_parser_node *node = json_parser_top(p);
1178     if (node->json->type == JSON_OBJECT) {
1179         json_object_put(node->json, p->member_name, value);
1180         free(p->member_name);
1181         p->member_name = NULL;
1182     } else if (node->json->type == JSON_ARRAY) {
1183         json_array_add(node->json, value);
1184     } else {
1185         NOT_REACHED();
1186     }
1187 }
1188
1189 static struct json_parser_node *
1190 json_parser_push(struct json_parser *p,
1191                  struct json *new_json, enum json_parse_state new_state)
1192 {
1193     if (p->height < JSON_MAX_HEIGHT) {
1194         struct json_parser_node *node;
1195
1196         if (p->height >= p->allocated_height) {
1197             p->stack = x2nrealloc(p->stack, &p->allocated_height,
1198                                   sizeof *p->stack);
1199         }
1200
1201         if (p->height > 0) {
1202             json_parser_put_value(p, new_json);
1203         }
1204
1205         node = &p->stack[p->height++];
1206         node->json = new_json;
1207         p->parse_state = new_state;
1208         return node;
1209     } else {
1210         json_destroy(new_json);
1211         json_error(p, "input exceeds maximum nesting depth %d",
1212                    JSON_MAX_HEIGHT);
1213         return NULL;
1214     }
1215 }
1216
1217 static void
1218 json_parser_push_object(struct json_parser *p)
1219 {
1220     json_parser_push(p, json_object_create(), JSON_PARSE_OBJECT_INIT);
1221 }
1222
1223 static void
1224 json_parser_push_array(struct json_parser *p)
1225 {
1226     json_parser_push(p, json_array_create_empty(), JSON_PARSE_ARRAY_INIT);
1227 }
1228
1229 static void
1230 json_parse_value(struct json_parser *p, struct json_token *token,
1231                  enum json_parse_state next_state)
1232 {
1233     struct json *value;
1234
1235     switch (token->type) {
1236     case T_FALSE:
1237         value = json_boolean_create(false);
1238         break;
1239
1240     case T_NULL:
1241         value = json_null_create();
1242         break;
1243
1244     case T_TRUE:
1245         value = json_boolean_create(true);
1246         break;
1247
1248     case '{':
1249         json_parser_push_object(p);
1250         return;
1251
1252     case '[':
1253         json_parser_push_array(p);
1254         return;
1255
1256     case T_INTEGER:
1257         value = json_integer_create(token->u.integer);
1258         break;
1259
1260     case T_REAL:
1261         value = json_real_create(token->u.real);
1262         break;
1263
1264     case T_STRING:
1265         value = json_string_create(token->u.string);
1266         break;
1267
1268     case T_EOF:
1269     case '}':
1270     case ']':
1271     case ':':
1272     case ',':
1273     default:
1274         json_error(p, "syntax error expecting value");
1275         return;
1276     }
1277
1278     json_parser_put_value(p, value);
1279     p->parse_state = next_state;
1280 }
1281
1282 static void
1283 json_parser_pop(struct json_parser *p)
1284 {
1285     struct json_parser_node *node;
1286
1287     /* Conserve memory. */
1288     node = json_parser_top(p);
1289     if (node->json->type == JSON_ARRAY) {
1290         json_array_trim(node->json);
1291     }
1292
1293     /* Pop off the top-of-stack. */
1294     if (p->height == 1) {
1295         p->parse_state = JSON_PARSE_END;
1296         if (!(p->flags & JSPF_TRAILER)) {
1297             p->done = true;
1298         }
1299     } else {
1300         p->height--;
1301         node = json_parser_top(p);
1302         if (node->json->type == JSON_ARRAY) {
1303             p->parse_state = JSON_PARSE_ARRAY_NEXT;
1304         } else if (node->json->type == JSON_OBJECT) {
1305             p->parse_state = JSON_PARSE_OBJECT_NEXT;
1306         } else {
1307             NOT_REACHED();
1308         }
1309     }
1310 }
1311
1312 static void
1313 json_parser_input(struct json_parser *p, struct json_token *token)
1314 {
1315     switch (p->parse_state) {
1316     case JSON_PARSE_START:
1317         if (token->type == '{') {
1318             json_parser_push_object(p);
1319         } else if (token->type == '[') {
1320             json_parser_push_array(p);
1321         } else {
1322             json_error(p, "syntax error at beginning of input");
1323         }
1324         break;
1325
1326     case JSON_PARSE_END:
1327         json_error(p, "trailing garbage at end of input");
1328         break;
1329
1330     case JSON_PARSE_OBJECT_INIT:
1331         if (token->type == '}') {
1332             json_parser_pop(p);
1333             break;
1334         }
1335         /* Fall through. */
1336     case JSON_PARSE_OBJECT_NAME:
1337         if (token->type == T_STRING) {
1338             p->member_name = xstrdup(token->u.string);
1339             p->parse_state = JSON_PARSE_OBJECT_COLON;
1340         } else {
1341             json_error(p, "syntax error parsing object expecting string");
1342         }
1343         break;
1344
1345     case JSON_PARSE_OBJECT_COLON:
1346         if (token->type == ':') {
1347             p->parse_state = JSON_PARSE_OBJECT_VALUE;
1348         } else {
1349             json_error(p, "syntax error parsing object expecting ':'");
1350         }
1351         break;
1352
1353     case JSON_PARSE_OBJECT_VALUE:
1354         json_parse_value(p, token, JSON_PARSE_OBJECT_NEXT);
1355         break;
1356
1357     case JSON_PARSE_OBJECT_NEXT:
1358         if (token->type == ',') {
1359             p->parse_state = JSON_PARSE_OBJECT_NAME;
1360         } else if (token->type == '}') {
1361             json_parser_pop(p);
1362         } else {
1363             json_error(p, "syntax error expecting '}' or ','");
1364         }
1365         break;
1366
1367     case JSON_PARSE_ARRAY_INIT:
1368         if (token->type == ']') {
1369             json_parser_pop(p);
1370             break;
1371         }
1372         /* Fall through. */
1373     case JSON_PARSE_ARRAY_VALUE:
1374         json_parse_value(p, token, JSON_PARSE_ARRAY_NEXT);
1375         break;
1376
1377     case JSON_PARSE_ARRAY_NEXT:
1378         if (token->type == ',') {
1379             p->parse_state = JSON_PARSE_ARRAY_VALUE;
1380         } else if (token->type == ']') {
1381             json_parser_pop(p);
1382         } else {
1383             json_error(p, "syntax error expecting ']' or ','");
1384         }
1385         break;
1386
1387     default:
1388         abort();
1389     }
1390
1391     p->lex_state = JSON_LEX_START;
1392     ds_clear(&p->buffer);
1393 }
1394
1395 static struct json *
1396 json_create(enum json_type type)
1397 {
1398     struct json *json = xmalloc(sizeof *json);
1399     json->type = type;
1400     return json;
1401 }
1402
1403 static void
1404 json_error(struct json_parser *p, const char *format, ...)
1405 {
1406     if (!p->error) {
1407         struct ds msg;
1408         va_list args;
1409
1410         ds_init(&msg);
1411         ds_put_format(&msg, "line %d, column %d, byte %d: ",
1412                       p->line_number, p->column_number, p->byte_number);
1413         va_start(args, format);
1414         ds_put_format_valist(&msg, format, args);
1415         va_end(args);
1416
1417         p->error = ds_steal_cstr(&msg);
1418
1419         p->done = true;
1420     }
1421 }
1422 \f
1423 #define SPACES_PER_LEVEL 2
1424
1425 struct json_serializer {
1426     struct ds *ds;
1427     int depth;
1428     int flags;
1429 };
1430
1431 static void json_serialize(const struct json *, struct json_serializer *);
1432 static void json_serialize_object(const struct shash *object,
1433                                   struct json_serializer *);
1434 static void json_serialize_array(const struct json_array *,
1435                                  struct json_serializer *);
1436 static void json_serialize_string(const char *, struct ds *);
1437
1438 /* Converts 'json' to a string in JSON format, encoded in UTF-8, and returns
1439  * that string.  The caller is responsible for freeing the returned string,
1440  * with free(), when it is no longer needed.
1441  *
1442  * If 'flags' contains JSSF_PRETTY, the output is pretty-printed with each
1443  * nesting level introducing an additional indentation.  Otherwise, the
1444  * returned string does not contain any new-line characters.
1445  *
1446  * If 'flags' contains JSSF_SORT, members of objects in the output are sorted
1447  * in bytewise lexicographic order for reproducibility.  Otherwise, members of
1448  * objects are output in an indeterminate order.
1449  *
1450  * The returned string is valid JSON only if 'json' represents an array or an
1451  * object, since a bare literal does not satisfy the JSON grammar. */
1452 char *
1453 json_to_string(const struct json *json, int flags)
1454 {
1455     struct ds ds;
1456
1457     ds_init(&ds);
1458     json_to_ds(json, flags, &ds);
1459     return ds_steal_cstr(&ds);
1460 }
1461
1462 /* Same as json_to_string(), but the output is appended to 'ds'. */
1463 void
1464 json_to_ds(const struct json *json, int flags, struct ds *ds)
1465 {
1466     struct json_serializer s;
1467
1468     s.ds = ds;
1469     s.depth = 0;
1470     s.flags = flags;
1471     json_serialize(json, &s);
1472 }
1473
1474 static void
1475 json_serialize(const struct json *json, struct json_serializer *s)
1476 {
1477     struct ds *ds = s->ds;
1478
1479     switch (json->type) {
1480     case JSON_NULL:
1481         ds_put_cstr(ds, "null");
1482         break;
1483
1484     case JSON_FALSE:
1485         ds_put_cstr(ds, "false");
1486         break;
1487
1488     case JSON_TRUE:
1489         ds_put_cstr(ds, "true");
1490         break;
1491
1492     case JSON_OBJECT:
1493         json_serialize_object(json->u.object, s);
1494         break;
1495
1496     case JSON_ARRAY:
1497         json_serialize_array(&json->u.array, s);
1498         break;
1499
1500     case JSON_INTEGER:
1501         ds_put_format(ds, "%lld", json->u.integer);
1502         break;
1503
1504     case JSON_REAL:
1505         ds_put_format(ds, "%.*g", DBL_DIG, json->u.real);
1506         break;
1507
1508     case JSON_STRING:
1509         json_serialize_string(json->u.string, ds);
1510         break;
1511
1512     case JSON_N_TYPES:
1513     default:
1514         NOT_REACHED();
1515     }
1516 }
1517
1518 static void
1519 indent_line(struct json_serializer *s)
1520 {
1521     if (s->flags & JSSF_PRETTY) {
1522         ds_put_char(s->ds, '\n');
1523         ds_put_char_multiple(s->ds, ' ', SPACES_PER_LEVEL * s->depth);
1524     }
1525 }
1526
1527 static void
1528 json_serialize_object_member(size_t i, const struct shash_node *node,
1529                              struct json_serializer *s)
1530 {
1531     struct ds *ds = s->ds;
1532
1533     if (i) {
1534         ds_put_char(ds, ',');
1535         indent_line(s);
1536     }
1537
1538     json_serialize_string(node->name, ds);
1539     ds_put_char(ds, ':');
1540     if (s->flags & JSSF_PRETTY) {
1541         ds_put_char(ds, ' ');
1542     }
1543     json_serialize(node->data, s);
1544 }
1545
1546 static void
1547 json_serialize_object(const struct shash *object, struct json_serializer *s)
1548 {
1549     struct ds *ds = s->ds;
1550
1551     ds_put_char(ds, '{');
1552
1553     s->depth++;
1554     indent_line(s);
1555
1556     if (s->flags & JSSF_SORT) {
1557         const struct shash_node **nodes;
1558         size_t n, i;
1559
1560         nodes = shash_sort(object);
1561         n = shash_count(object);
1562         for (i = 0; i < n; i++) {
1563             json_serialize_object_member(i, nodes[i], s);
1564         }
1565         free(nodes);
1566     } else {
1567         struct shash_node *node;
1568         size_t i;
1569
1570         i = 0;
1571         SHASH_FOR_EACH (node, object) {
1572             json_serialize_object_member(i++, node, s);
1573         }
1574     }
1575
1576     ds_put_char(ds, '}');
1577     s->depth--;
1578 }
1579
1580 static void
1581 json_serialize_array(const struct json_array *array, struct json_serializer *s)
1582 {
1583     struct ds *ds = s->ds;
1584     size_t i;
1585
1586     ds_put_char(ds, '[');
1587     s->depth++;
1588
1589     if (array->n > 0) {
1590         indent_line(s);
1591
1592         for (i = 0; i < array->n; i++) {
1593             if (i) {
1594                 ds_put_char(ds, ',');
1595                 indent_line(s);
1596             }
1597             json_serialize(array->elems[i], s);
1598         }
1599     }
1600
1601     s->depth--;
1602     ds_put_char(ds, ']');
1603 }
1604
1605 static void
1606 json_serialize_string(const char *string, struct ds *ds)
1607 {
1608     uint8_t c;
1609
1610     ds_put_char(ds, '"');
1611     while ((c = *string++) != '\0') {
1612         switch (c) {
1613         case '"':
1614             ds_put_cstr(ds, "\\\"");
1615             break;
1616
1617         case '\\':
1618             ds_put_cstr(ds, "\\\\");
1619             break;
1620
1621         case '\b':
1622             ds_put_cstr(ds, "\\b");
1623             break;
1624
1625         case '\f':
1626             ds_put_cstr(ds, "\\f");
1627             break;
1628
1629         case '\n':
1630             ds_put_cstr(ds, "\\n");
1631             break;
1632
1633         case '\r':
1634             ds_put_cstr(ds, "\\r");
1635             break;
1636
1637         case '\t':
1638             ds_put_cstr(ds, "\\t");
1639             break;
1640
1641         default:
1642             if (c >= 32) {
1643                 ds_put_char(ds, c);
1644             } else {
1645                 ds_put_format(ds, "\\u%04x", c);
1646             }
1647             break;
1648         }
1649     }
1650     ds_put_char(ds, '"');
1651 }